Quaresmi Seminar on Monday 13 May 2024 : “Challenges in Modelling Stochastic Processes with Queueing Systems: From Statistical Inference to Burstiness Analysis”

Monday 13 May 2024 at 1pm we will welcome Azam Asanjarani, Senior Lecturer at the University of Auckland. She will present her talk “Challenges in Modelling Stochastic Processes with Queueing Systems: From Statistical Inference to Burstiness Analysis”.

In this talk, we explore the interface of stochastic modelling and statistical inference, with a focus on the challenges in modelling stochastic processes with queueing systems. The evolution of queuing systems often happens randomly and key system variables may be unknown or only partially observed. The ideal system is a stable system that reduces the customers’ waiting time or makes it stochastically bounded. To achieve system stability we need to estimate parameters and typical performance measures of the system. Evaluating these performance measures allows major improvements in the design of new systems, modifications made to existing systems, forecasting and online control. However, applying queueing theory results to real-world systems raises questions about statistical analysis, such as “How would we measure parameters?” or “Are the assumptions about the model sensible and supported by data?”. To answer these questions we need to investigate inference methods for queueing systems corresponding to our observed data. The next problem that we might be confronted with when modelling real-life phenomena is the burstiness of the arrival process. Even in a stable queueing system, due to the random nature of the system, having many arrivals into the system within a short duration of time can make the system unstable. Stochastic processes based on Markov models are very popular for modelling arrival or service processes where the dependence on the history of the process is negligible. Here, we explore the properties of bursty Markovian arrival processes (MAPs) and establish the mathematical properties of “bursty” processes.

The seminar will take place in room C001 at Anjou campus (365A, Rue au Bois,1150 Brussels). Please register at quaresmi@ichec.be